Node Behavior Classification for Traffic Prediction in Optical Burst Switched Networks using Machine Learning
Deepali Bhawarthi1, Girish Chowdhary2

1Deepali Bhawarthi*, Department of Computer Engineering, AISSMS College of Engineering, Pune, India.
2Dr Girish Chowdhary, School of Computational Sciences, SRTMU, Nanded, India.
Manuscript received on January 14, 2020. | Revised Manuscript received on January 26, 2020. | Manuscript published on February 10, 2020. | PP: 1963-1968 | Volume-9 Issue-4, February 2020. | Retrieval Number: D1305029420/2020©BEIESP | DOI: 10.35940/ijitee.D1305.029420
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: Currently due to massive use of internet there is need of huge amount of bandwidth. The utilization of bandwidth can be managed up with optical burst switched networks. These networks cannot provide good QoS due to problems like wavelength contention and congestion problem. Also it is not necessary that contention in a network leads to congestion. It can be due to nodes behavior which affects the flow of traffic from source to destination. Hence there is a need to classify the traffic through the node at correct juncture to avoid congestion. This can be achieved using machine learning techniques. In this paper, support vector machine, AdaBoost classifier and Bagging classifier are evaluated .Experimental work is carried on Optical Burst Switched network dataset using 22 attributes which is available on UCI repository. The results show that bagging classifier performed better with accuracy of 95% in classifying the nodes behavior. 
Keywords: Optical Burst Switched Networks, Machine learning methods, Support vector machine, AdaBoost, Bagging
Scope of the Article: Machine learning